-
CENTRES
Progammes & Centres
Location
Artificial Intelligence must be deployed in OTNs to ensure scalable and equitable digital participation across socio-economic segments
Image Source: Getty
The global community is at a critical inflection point marked by the simultaneous emergence of two distinct transformative technological paradigms: Artificial Intelligence (AI) and Digital Public Infrastructure (DPI). Their convergence presents a unique opportunity to catalyse unprecedented, non-linear growth across countries.
Traditionally, DPIs were societal in nature, funded and operated by governments. Examples include digital IDs, payment systems, and data exchange platforms providing services to citizens and businesses. DPIs are now evolving into a new category termed “Market DPIs”. Market DPIs aim to establish equitable, accessible, and democratic markets through open, interoperable frameworks. These frameworks mitigate monopolistic dominance by enabling value creation across sectors. Market DPIs are bringing together organisations, governments, and consumers by triggering a flywheel effect of innovation and advancing their respective objectives. Unlike societal DPIs, with market DPIs, governments act more as facilitators—instituting enabling policies while refraining from owning intellectual property or directly funding technology systems using taxpayers' money.
Open Transaction Networks (OTNs), a subset of Market DPIs, exemplify this shift from platform-centric to protocol-driven commerce. OTNs, such as the Open Network for Digital Commerce (ONDC), Digital Energy Grid (DEG), Open Agri Net (OAN), and similar initiatives across the globe are redefining digital transactions by promoting open standards over centralised control and enabling decentralised and participatory digital commerce. When enhanced with AI, these networks can democratise commerce, enhance operational efficiencies, and foster innovation across the value chain. OTNs provide rich and transparent datasets, offering comprehensive feedback to refine underlying AI models, mitigate biases, and enhance efficacy.
OTNs are inherently designed to prioritise universal inclusivity and accessibility, ensuring equitable service delivery to all residents, including those in the most remote and underserved communities.
Unlike traditional digital commerce platforms that operate as closed ecosystems with proprietary rules and infrastructure, OTNs function on open and standards-based protocols that allow diverse participants to connect and transact without intermediary control. This fosters innovation and competition by ensuring that no single entity holds disproportionate influence over the market. Any catalogable product or service, or a combination thereof, falls within the purview of digital commerce.
Open networks exhibit the following features:
Interoperability: Standardised protocols that enable diverse systems to communicate seamlessly
Decentralisation: Reduced reliance on central intermediaries or dominant platforms
Open protocols: Publicly available specifications that allow anyone to build compatible solutions
Network effects: Value increases when a new organisation, whether large or small, joins the ecosystem, thus amplifying collective benefits.
Figure 1: Use cases of OTNs
A diverse range of OTNs is emerging, with some focused on specific sectors such as agriculture or energy, and others spanning multiple domains such as digital commerce or tailored to national priorities. Each is designed to address unique local challenges, promote inclusive digital participation, or harness a new opportunity. While these networks are currently not interoperable, they may, in due course, be modelled and evolve like the internet. As OTNs begin to be standardised and interoperated across national boundaries, they can unlock massive economic value through seamless, protocol-based trade and collaboration on a global scale. This is especially true for Medium, Small and Micro Enterprises (MSMEs) that are often excluded from cross-border markets due to cost, complexity and lack of visibility. With interoperability as the fulcrum, OTNs can also power intercountry knowledge grids, opening up frontiers for skills exchange.
Much like the roads and other public infrastructure, OTNs provide a digital highway for transactions, improving the supply and quality of training and post-training data for AI models.
OTNs are inherently designed to prioritise universal inclusivity and accessibility, ensuring equitable service delivery to all residents, including those in the most remote and underserved communities. AI serves as a critical enabler in this endeavour by addressing linguistic divides through vernacular language support. Beyond mere translation, AI’s advanced reasoning capabilities dynamically generate contextually relevant insights, thereby reducing information asymmetry and empowering users with actionable and localised knowledge. This synergy aligns well with the principles of universal access and social inclusion.
The penetration of digital commerce in most countries, especially in the Global South, remains fairly limited despite widespread connectivity, smartphone adoption, and a young, aspirational population.This is primarily due to language barriers and information asymmetry.
AI interventions can improve efficiencies for both buyers and sellers. They can not only increase adoption but also enable a larger cross-section to participate in the formal economy. These interventions can include:
Multi-modal, multi-language interfaces (Buyer side): Voice-enabled buyer apps, powered by technologies that offer real-time translation, transliteration, automatic speech recognition, text-to-speech and speech-to-text services can support interactions across various national and international languages. Buyers will be able to complete transaction activities—from search to payment and post-order transactions - with a hybrid interface incorporating voice, text, visuals, and even gestures for specific use-cases. Buyers will have the flexibility to switch among the different inputs seamlessly. Contextualised search can be enabled in regional languages and dialects. Product and service descriptions can be made available in local dialects, assisting buyers in making an informed choice. The conversation between the buyer and the buyer app can support mixed language modes—for example Hindi and English, French and English—while also accommodating idioms and slangs in the local languages, making these interactions more natural and realistic.
Personalisation and recommendation engines (Buyer side): AI can facilitate personalisation for buyers and sellers through recommendation engines, which are a class of algorithms that can personalise content recommendations based on user preferences. These recommendation engines can help in upselling and cross-selling of products and services while adhering to the data privacy regulations of the country.
Catalogue creation and management (Seller side): One of the bottlenecks for small businesses, nano entrepreneurs and artisans has been the creation and maintenance of good quality catalogues of their products and services that comply with legal metrology standards while providing a compelling experience for buyers. By leveraging AI and computer vision, small businesses can create professional-grade catalogues with user-friendly tools in a short period. AI-driven cataloguing can help standardise the product experience by ensuring compliance, accuracy and completeness of information, thereby reducing errors and inconsistencies in listings.
Risk assessment models (Seller side): The rich availability of transaction data can enable more sophisticated approaches to business risk assessment. Financial institutions could develop better lending models based on actual business performance rather than traditional metrics, using alternate data to construct underwriting algorithms. This could transform small businesses that lack a verifiable credit history, enabling them to access loans at market rates. Making data-driven decision-making accessible to millions of small businesses could help unlock the efficiency and growth potential of the country's vast informal sector.
Countries that invest in AI-ready OTNs can ensure that AI serves public interest while driving innovation. However, challenges such as decentralised AI development, large training and inference costs, data privacy concerns, regulatory challenges, and concerns regarding transparency and fairness must be addressed collaboratively by AI service providers and policymakers. This must be done to ensure inclusivity and accessibility without compromising consumer rights.
Historically, technological systems functioned within deterministic and predictable frameworks. The emergence of AI, however, represents a paradigm shift, which demands both precision in AI models and rigorous adherence to the principles of fairness and transparency.
Much like the roads and other public infrastructure, OTNs provide a digital highway for transactions, improving the supply and quality of training and post-training data for AI models. Real-world data is a key input for modern AI systems, and OTNs have the unique potential to significantly enhance its availability. There are substantial concerns that Large Language Models (LLMs) rely heavily on synthetic data for extensive training, which poses significant risks to model reliability and may lead to performance degradation or collapse. OTNs, like digital ID and digital payment, collect large amounts of data from buyers and sellers after obtaining informed consent. This ethically sourced, consent-based data can significantly enhance the quality of training and post-training datasets for frontier AI models by providing diverse, real-world inputs. Unlike synthetic data, which risks compounding inaccuracies or biases, human-generated data from OTNs is grounded in actual behaviour and transactions. As a result, OTNs can help overcome the ‘data walls’ impeding AI development, enabling the creation of more accurate, robust, and inclusive frontier AI systems.
Over a period of time, OTNs can provide structured and standardised datasets that could potentially lead to significant gains in model performance. But beyond data standardisation, they address a deeper systemic issue in AI development - ‘Algorithmic Bias’, one of the key impediments to building trust.
Marginalised populations are often underrepresented in AI datasets, leading to algorithmic outcomes that are inaccurate or even harmful to these communities. Most importantly, OTNs can formalise an ethical way of collecting the data through a consent-driven mechanism, thus diversifying datasets and embedding transparency, fairness and equity at the foundation of AI systems.
In a nutshell, the convergence of AI and OTNs indicates a paradigm shift toward accessibility, inclusivity, and economic democratisation. Their mutually reinforcing relationship—AI enhancing OTNs, and OTNs refining AI—promises to reshape global digital ecosystems.
Enabling AI for OTNs requires an approach rooted in responsibility, inclusivity, and contextual awareness. This demands responsible AI practices, compliance with data privacy regulations, the establishment of robust governance frameworks, and local language contextualisation to align with the digital literacy levels of users and the maturity of a country’s digital infrastructure. Equally important is the issue of cost: for AI to operate at the population scale, OTNs must be democratised and made affordable, to encourage use across diverse socio-economic segments. Realising this vision will require active participation from the market ecosystem, policy advocates and funding institutions to ensure AI deployment in OTNs is socially aligned and scalable.
Neeraj Jain is the Vice President of Network Enhancement at Open Network For Digital Commerce
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.